The nature of poverty is complex, and its causes are diverse; therefore intervention to reduce poverty must be tailored to particular circumstances. There are at least four strategies adopted to assist households exit from poverty: agricultural intensification and diversification, increased farm size, increased off-farm employment and income in rural areas, and rural-urban migration or exit from agriculture (Dixon et al., 2001; ADB, 2006). Farmland is the most important factor in exiting poverty as the agricultural sector is traditionally the dominant employer. Access to farmland is more crucial when off-farm employment and migration are not available.
Land policy is important but complex as it is country-specific, long-term in nature and politically controversial (Deininger, 2003). Land is a key asset for the poor, and provides a foundation for economic activities and non-market institutions. Land policy addresses structural issues through improving economic opportunities for the poor. The challenges for land policy options are: giving stronger rights to disadvantaged people; allowing transferability of land; and encouraging a rental land market for wide-ranging productive economic activities. To make land reform policy feasible and applicable, it should be carefully managed, supported and co-ordinated by high-level governmental institutions.
Based on this background a study on land economy was initiated. It aimed to analyse the status of the land economy in the last two decades using the data from the Indonesian agricultural census from 1983-2003, and to establish the future direction of land policy dealing with the problems faced by marginal farmers in improving farm household income and welfare.
Current status of land economy
This section consists of four inter-related subjects including agricultural land conversion, agricultural land distribution, small farm household and land size, and the land-derived agricultural income required to support household needs.
Increases in the proportion of small-scale farmers would have an impact on the land distribution gap and could lead to rapid agricultural land conversion. For farmers with restricted off-farm employment, limited access to land will seriously affect the capacity of agricultural income to support household expenditure; and this leads to a decrease in the farmer's welfare.
Land conversion and land use planning
In the first decade (1983 to 1993), the conversion of agricultural land reached 1.280 million hectares (Table 1). The majortiy was in Java where it reached 1.015 million ha or 79.3 per cent, and most of this (68.3 per cent) was in agricultural wetland (Anonymous, 1996). In the next decade, 1993 to 2003, there was insignificant change in the total amount of land conversion, 1.264 million ha mainly executed in Sumatra (92.3 per cent). According to Nasoetion (2004), the total agricultural wetland conversion in Java for the period 1998 to 2004 increased, 23,600 ha per year or 61 per cent of the total converted land in the region.
The proportion of land conversion with respect to the total agricultural land in these two decades increased from 7.7 per cent to 8.2 per cent. These results indicate that during the last decade, 1993 to 2003, the conversion rate of agricultural land outside Java was relatively high; for instance in Sumatra it was 21.5 per cent, Papua 19.2 per cent and Maluku islands 12.1 per cent. However the conversion ratio in Java decreased from 18.7 to 8.8 per cent (Table 1). The widening trend of agricultural land conversion is alarming for planners and practitioners involved in agricultural development because it takes place on productive agricultural land and/or in regions that had good agriculture and rural infrastructure, and also because of the high cost of new agricultural land development outside Java.
Analysis of agricultural land conversion should be complemented with the land utilization data of the national balance sheet. In the National Land Use Planning that was proposed by the National Land Agency (Badan Pertanahan Nasional/BPN) it was pointed out that the land allocated for environmental sustainability is 191 million ha including an allocation of 35.4 per cent for a 'conservation zone' and 64.6 per cent for a 'cultivation zone' (BPN, 2001 in Silalahi, 2006). In reality, 18.4 per cent or 12 million ha of the land in the conservation area has been exploited while 57.7 per cent or 71 million ha in the cultivation zone has not yet been developed.
Agricultural landholding distribution
The Gini ratio index used to determine the distribution gap of landholdings, ranges from 0 to 1.0. The index value indicates the width of the distribution gap, with a threshold level of 0.50 (less than 0.5 indicates the gap is low, greater than 0.5 indicates the gap is high) (Oshima, 1976 in Rusastra and Sudaryanto, 1999). In the three decades from 1973 to 2003 the land distribution gap consistently widened, with the Gini index increasing from 0.5481 to 0.7171 (Table 2). The distribution gap of landholdings was determined by the comparison of the Gini ratio of the household land size. A small average landholding (less than 0.10 ha) is a source of distribution inequality. The land distribution gap in Java, that generally started in 1993, was wider than outside Java, which started in 2003 (Gini ratio > 0.50).
Rusastra et al. (2007) provide figures on the Gini ratio of the land distribution by type for the period of 1993-2003. These are: (a) the Gini ratio of wetland versus dry land in Java is relatively similar, and there is no improvement over time; (b) the Gini ratio of wetland outside Java significantly decreased from 0.7154 to 0.4784 while the dry land remained constant at 0.5700; (c) even though, in 2003 the wetland holding distribution outside Java improved, its Gini ratio was 0.4784, approaching the threshold level 0.50; (d) in general, the distribution gap of wetland and dry land in Java and off-Java was high, as its Gini ratio was greater than 0.50. There is no clear indication whether agriculture wetland with higher productivity will have a higher Gini ratio index. Both wetland and dry land markets have great demand and the scarcity of land is relatively similar.
Small farmer and land size
A descriptive analysis on the growth of three categories, agricultural household, household landholding, and small farm household with landholding less than 0.50 ha, is presented in Table 3. In the last decade, the national proportion of the household landholding (with respect to farm household) decreased from 98.71 per cent to 95.22 per cent while the percentage of small farm household increased from 52.66 per cent to 56.20 per cent. The annual growth rate of the small farm household was higher than the household landholding: 2.39 per cent versus 1.73 per cent per year. The population of marginal farmers with land size less than 0.50 ha substantially increased from 10.8 million to 13.7 million households.
In Java, the existence of marginal farmers was more apparent than outside Java, the proportion increased from 69.76 per cent to 74.68 per cent with an annual growth rate of 2.16 per cent per year. Although its percentage in off-Java is lower, the growth was faster with the rate 3.05 per cent per year. In 2003, the populations of these groups in and outside Java were 9.99 million and 3.70 million. For farmers with limited availability and access to off-farm employment and income, agricultural-derived income will not be sufficient to fulfil their household needs. Most of them are sensitive to external economic shocks and frequently experience vulnerability and food insecurity in their life.
Rusastra et al. (2007) also illustrated the descriptive analysis on food crop household by land size category for the period of 1983-2003 that provides a strong indication of land fragmentation. These indications are: (a) the average of landholding at the national level decreased from 0.99 to 0.79 ha; (b) the condition of the farmers with land size 0.10�0.49 ha: the occupation of agricultural land reduced from 10.12 to 4.46 per cent and the average landholding decreased from 0.27 to 0.09 ha; (c) the status of the farmers with land size more than 2.0 ha: the population less than 14.9 per cent, the occupation of agricultural land is approaching half of the total land while the trend of the total occupation decreased from 49.47 to 46.41 per cent, and the land size average decreased from 3.62 to 3.26 ha; and (d) in contrast, the population of the group of farm households with land size less than 2.0 ha was around 89.0 per cent in 2003, the land occupation only 54.0 per cent of the total agricultural land. In general, this evidence indicates the inequality in land distribution amongst the farmer groups; in addition, serious land fragmentation would lead to poverty and insufficiency to support farmers' welfare.
The intervention of government and private sector in promoting small-scale farming as a driver of growth and poverty reduction should consider the current debates on the potential of small farms (Hazell et al., 2007). In terms of efficiency, small farms typically use land intensively and employ a lot of labour. Relating to equity and poverty reduction, small farms are preferred to large. On the other hand, external economic shocks affect both small and large farms. But other developments may pose more severe challenges for smallholdings. New technologies requiring more capital inputs, mechanization, or high levels of education, may disadvantage smaller farms. More worrying are the implications of changes to marketing chains. Supermarket operators are becoming increasingly important in parts of the developing world. Supermarkets impose very strict standards for the quality, consistency, and timeliness of supply.
The existence of agricultural land-derived income
The capacity of land-derived agricultural income to support the household needs is presented in Table 4. In the last two decades, 1983 to 2003, there was an indication of the decrease in the farmers' livelihood. The proportion of households who have sufficient farm-derived income dropped from 59.4 to 49.3 per cent while the percentage of the farm families who have insufficient or far from sufficient income substantially increased from 37.9 to 47.4 per cent. In 2003, the proportions of households in and outside Java that had insufficient or far from sufficient land-derived income to support their needs were 53.8 per cent and 39.9 per cent. This information specified that necessary efforts should be made to increase the households' income through land-based or non farm-derived income.
The household income structure in Indonesia for 2003 indicated that 60.5 per cent of income was from farm activities and 39.5 per cent was from non-agricultural sector (Rusastra et al., 2007). Balisacan et al. (2002) found that there were huge differences in poverty change, sub-national economic growth, and local attributes such as terms of trade regime, schooling, infrastructure and access to technology. The welfare of the poor responded quite strongly to overall income growth, in which the poverty growth elasticity is 0.7, due to the high contribution of agricultural growth. To improve the status of the welfare of the rural poor, a focus on agriculture and rural development is necessary while at the same time maintaining the growth of the non-agricultural sector.
Future policy direction
There are at least four agricultural land-related problems. For instance: (a) rapid land conversion to non-agricultural uses especially in productive agricultural land; (b) high inequality in land distribution in and outside Java; (c) the existence of land fragmentation, decreasing land size, significant increase in the proportion of marginal farmers; and (d) insufficient land-derived agricultural income to support household needs. The future policy for responding to these problems should be based on the previous descriptive analysis and current literature review on the relevant subject.
Strategic policies for preventing excessive land conversion should consider the following options (Kasryno et al., 1996, Pasandaran et al., 1998, and Silalahi et al., 2006): (a) to develop high-value agricultural commodities in Java due to the high demand for and scarcity of land; (b) to improve the existing capacity and utilization of irrigated land and irrigation infrastructure; (c) to establish and improve regional land planning and utilization; (d) to develop regulations and law enforcement, zoning and incentive systems; and (e) to develop new agricultural land outside Java.
Land policies dealing with the decreasing land size and inequality in agricultural land distribution need to consider the following options (Rusastra et al., 2007): (a) to support land consolidation especially in regions with improved infrastructure and access to non-agricultural employment; (b) to support land transferability (land rental) through developing agricultural commodities with better comparative advantage; (c) to encourage the internal land consolidation among farmers, complemented with improved access to part-time employment in urban areas; and (d) to stimulate and develop the agribusiness partnership between the private sector and farmer groups in rural areas.
Appropriate approaches for decreasing the number of marginal farmers and improving the agricultural land-derived income are basically related to the improvement of access of the small farmers to agricultural land. UNU/WIDER1 indicated that there are many alternative paths of access to land for marginal farmers, including formal and informal as well as spontaneous and heavily regulated pathways (Janvry and Sadoulet, 2001). The potential for community titling exists where communities have a sufficient level of social capital and leadership to manage the resource efficiently. Access to land through rental contracts (share arrangements in particular) can be effective under the condition of extensive market and institutional failures. On the other hand, formal registration and titling are essential if land is scarce and valuable, and local social capital is no longer sufficient to guarantee property rights and land trades.
In the case of Indonesia, there is no strong indication of land polarization, but land fragmentation does exist. In fact, 11.3 per cent of farm households occupied 46.4 per cent of agricultural land with an average land size of 3.3 ha per household. This size is appropriate and could reasonably lead to an increase in agricultural productivity, and support household needs. The proper strategy for promoting the marginal farmer's accessibility to land is to implement land distribution programmes (land allotment) instead of land redistribution. The evidence revealed that there are about 71 million ha of land in the cultivation zone that have not been exploited and can be developed and distributed for the benefit of marginal farmers.
The main problems faced by the farm households are high inequality of land distribution, decreasing land size, an increasing proportion of small farmers, and insufficient land-derived agricultural income to support household needs. Proposed policy options dealing with these issues are: (a) to give priority to agricultural and rural development, agricultural diversification, agricultural land consolidation, informal and formal activities in rural areas, and integration of rural-urban economies; (b) to develop and enhance a land tenure system (rental contracts or share arrangements) and informal land markets for better land accessibility for marginal farmers; and (c) to implement land distribution programmes that are complemented with effective agricultural and rural development.
The following arguments can be made both for and against small farms: efficiency, equity and poverty reduction, the application of new technology, and the existence of new marketing chains. Where there is no indication that supporting small farmers will have significant benefits, governments should encourage and support them to move into non-agricultural activities. Where small farms can provide benefits, the policy support should: (a) ensure a stable economy, provide public goods, good governance, and intervene in food and credit markets; (b) encourage farmers to follow market demand and improve their marketing; and (c) provide inputs and services to small farmers that are co-ordinated by all actors involved in agribusiness